Webinars · 9 min read

September 2026 Klient PSA Product Update Webinar

Watch the September 2026 Klient PSA webinar: how Klient rolls out the Klient PSA MCP and skills internally, a live demo of AI organizing and delivering a real project, and what's coming in 51.12.

Share

Not a typical feature month: with 51.12 arriving in October, Rosalie and Yanick used September for part two of our MCP series — how we roll out the Klient PSA MCP and skills across our own company, with a live demo of AI cleaning up and delivering a real project.

Webinar chapters

00:00 Welcome and agenda
02:54 Dreamforce takeaways
05:38 Next stop: Agentforce World Tour Amsterdam
06:26 MCP part 2 — recap of June
08:27 The Klient AI stack: agents, MCP and skills
11:03 The project as the core of AI delivery
12:11 Klient as customer zero
15:22 Demo setup: the Feature Hub project
20:26 Resource plan and the Workspace AI preset
23:34 Live demo: organizing a project with the Project skill
24:40 Live demo: turning a Slack channel into tasks
29:17 Q&A: managing token usage
34:02 PLANNY1 in Slack
37:48 How to get started: a four-step rollout
39:46 Community: Ideas Exchange, points and badges, eLearning
41:29 What’s coming in 51.12
43:02 Q&A and wrap-up

Dreamforce takeaways 02:54

The theme Yanick brought back from Dreamforce: let AI do the work instead of clicking through screens. Two announcements stood out. AIforce, a live interface where AI agents carry out tasks directly across Salesforce applications, and Claudeforce, which brings Salesforce directly into Claude. Both point the same way as Salesforce Headless 360: you reach your Salesforce data from wherever you work (Claude, Slack, agents), and the standard UI stops being the only way in.

That’s a direction Klient has been heading in for a while, with interfaces like the Workspace, Snap Workspace and the Resource Planner. It’s also exactly where MCP fits. If your Klient data is already in Salesforce, you’re in the right place.

Next stop is the Agentforce World Tour in Amsterdam on Thursday, November 26 (free, register here). Yanick, Rosalie, Raul and Krystel will be at the Klient booth, with meetings available on the Tuesday and Wednesday before. If you or your team will be nearby, reply to the recap email and come say hello.

The Klient AI stack today 08:27

This session picked up where our June Klient PSA MCP webinar left off (what MCP is, MCP vs agents, security). Three months later, here is what’s available:

  • AI agents that live in Salesforce and Slack — PLANNY1 for project management, TOUCHY1 for customer touchpoints and, new this month, TIMEY1, the timesheet agent, now officially available.
  • The Klient PSA MCP and skills, which live in your AI tool — Claude, Copilot, Cursor. MCP Data1 connects your AI to your Klient projects, tasks, time and assignments, and it cannot delete anything. Four skills sit on top: Project1 (build and update plans), Time1 (log time), Touchpoint1 (prep and follow up on customer calls, create tasks from them) and the newer Assignment1 (staff people and set schedules).
  • MCP Knowledge1 and the Knowledge1 skill, for creating and maintaining Salesforce Knowledge articles.
  • On the roadmap: SCOPEY1, STAFFY1 and MCP Admin1. Browse everything in the Feature Browser.

The project is the core of AI delivery 11:03

Yanick’s central point: when you deploy AI across a business, the project is the hub. It holds time, scope, milestones and actuals, it’s where you queue up the to-dos you want the AI to do, and it gives you the audit trail of what was done and by whom. The human leads, decides and approves; the AI does the doing. We call that Headless Project Delivery.

Klient is its own customer zero: 63 skills now run processes inside projects across software development, marketing, webinars, sales, support, customer success and time entry. Customer ideas from the community are scoped, developed, QA’d and documented, and documentation and comment updates are generated automatically. Marketing runs the website and ad budgets against objectives stored in a project so the AI doesn’t overspend. Rosalie’s webinar skills handle the work before and after every session like this one, from the invite and reminder list to the Zoom reports and recap. Everyone at Klient now updates the skills they use, so they get better every week.

Demo setup: a real project, an AI preset 15:22

The demo used a real internal project, the ongoing improvement of the Feature Hub, our customer community. It could just as easily be a billable customer project, the kind Salesforce partners deliver every day. The setup: Claude Code in VS Code, MCP Data1 and Knowledge1, the Project skill, and a Slack channel linked to the Salesforce project record (Klient names these AI-run projects with an “A-” prefix).

In the Resource Planner, the project looks simple: a few human tasks with Rosalie and Yanick booked roughly six hours a week. In the Workspace, Yanick uses a dedicated preset for AI projects, because the columns that matter change. Duration and dates matter less when a task takes minutes; status, next step, description, proposed solution and priority matter more. Moving a task to Ready for Dev is the signal that the AI can pick it up, so the human stays in control of what gets worked on. Tasks, and assigning them between humans and agents, are what make this Salesforce project management workable with AI.

Live demo: organizing a messy project 23:34

The Feature Hub project had a lot of tasks with no parent — a problem most teams recognize. Yanick asked the Project skill to organize it: give every task a parent, reuse existing parents or create new ones, and show the plan first. It proposed three new parent tasks (the store, gamification and adoption, and doc skills and tooling), mapped the loose tasks into them, and flagged the placements it was least sure about. Rosalie’s tip: review those flagged ones first. Yanick approved, and the reparenting ran through MCP Data1 while the demo carried on.

Live demo: from a Slack channel to delivered tasks 24:40

In a second thread running in parallel, Yanick asked Claude to read the project’s Slack channel, a messy stream of ideas and screenshots, and propose new tasks. It suggested three (two for an adoption dashboard, one for gamification). He told it to create them without parenting them, react with an 👀 emoji on the Slack posts to show they were being worked on, then fix the easiest one, move it to Ready for QA, and add a ✅. Minutes later the new tasks were in the project and one was ready for Rosalie to test.

Slack is just one source. As Rosalie pointed out, you can do the same from the last week of call transcripts, your email, or Teams through Copilot. The goal is to keep every task in Klient instead of spread across tools.

PLANNY1 in Slack 34:02

Skills and agents fit different moments. At your desk you might use Claude; on your phone, in Slack, you can invite PLANNY1 into the project channel and ask how many tasks are open (94, as it happened), then follow up with overdue tasks or status questions. Next release, the routines Yanick triggers by hand today will run on a schedule, the same way PLANNY1 already sends its 8 AM message.

How to get started: a four-step rollout 37:48

  1. Pick your team. Choose one or two reliable champions and get early wins before opening it up to everyone.
  2. Get the MCP and skills from My Products on the Feature Hub, then follow the install guide. We’re here to help.
  3. Create your first AI project in Klient, using the Project skill rather than building it by hand. That’s where your team’s AI work lives and where you’ll see everything that was done.
  4. Execute the work from your AI tool, whether that’s Claude, Copilot or Gemini.

Over time, share skills across the team and make each person responsible for improving their own.

Community: Ideas Exchange, points and badges, eLearning 39:46

Here’s what’s new on the Feature Hub, our customer community, since the last webinar:

  • Ideas Exchange: 129 ideas submitted since it launched in March, and 46 of them are already in development or live, more than a third. 14 new ideas came in over the last 30 days. The more votes an idea gets, the higher it moves, so go vote for the ones that matter to you.
  • Points and badges: you now earn points and badges for what you do on the Feature Hub, like completing a course, voting on ideas or attending webinars. Find yours, and your team’s, under My Profile.
  • eLearning leaderboard: track learning milestones and see who’s leading.
  • Self-registration: anyone on your team can create their own Feature Hub account from the login page. Just click “Not a member”.

What’s coming in 51.12 41:29

The current version is 51.11, released August 31. 51.12 is planned for October 9 and will be covered in full at the October webinar. Rosalie’s picks:

  • Draft mode in the Resource Planner — prepare assignment scenarios, then publish or discard them.
  • A Predecessor column — type dependencies instead of dragging them.
  • Group by Billable in My Timesheets.
  • Presets in Snap Workspace.
  • Survey Insights (coming in 51.13) — a new interface for consuming survey results.

Get the slides

Want the deck from this session? We’ll email you the full September slides, including the AI stack, how we set up AI projects and the four-step rollout.

Free download

Download the webinar deck

Get the full slide deck from this session — we'll email it straight to your inbox.

PDF · free · no spam

Frequently asked questions

How do you manage or predict token usage with the MCP connection?

Internally, everyone at Klient is on a Claude Max plan, and usage fits within that budget, so we don’t track tokens closely yet. That’s changing: in a future release agents will log their time and tokens against tasks, the same way people log time, so the cost lands on the project and you can see your real margin or recharge it. For your own estimate, run the skill on your use case and ask your AI tool to calculate the token usage. It varies a lot by person and workflow.

Can the Klient PSA MCP delete my projects or tasks?

No. MCP Data1 can create and update but cannot delete tasks or projects. If you ask it to restructure a project, it will add the replacements and tell you which old items to delete yourself.

Can I build my own skills?

Yes, and we encourage it. Start from a Klient skill, adapt it to your org or your role, and save it as a new skill. Rosalie uses a version of the Time skill customized for her role.

Do I need Slack to work this way?

No. Slack was the source in the demo, but the same approach works from call transcripts, email, or Microsoft Teams through Copilot. Any source your AI tool can read can become tasks in Klient.

What’s the difference between the MCP and the AI agents?

The MCP and skills are flexible and live in your AI tool (Claude, Copilot, Cursor). The agents, such as PLANNY1, TOUCHY1 and TIMEY1, are more guided and live inside Salesforce and Slack. Many teams use both, depending on where they are working at the time.

Does Klient see or log our prompts or data?

No. The MCP runs under your own Salesforce login and permissions, and Klient logs nothing. Your AI provider’s terms apply.

Which AI tools does it work with?

Claude works best. Copilot (including Copilot Studio), Cursor and Gemini CLI also work.

Join us next month

The next product update webinar is October 22, covering 51.12 in full. Register for the next Klient PSA webinar, and if you can’t join live the recording will be posted here afterwards.

Rosalie Chassat: Hi, everyone, and welcome to our September Product Update webinar! Wherever you’re joining from, thank you so much for being here. For those of you who don’t know me yet, I’m Rosalie, I’m in Customer Success at Klient, and I’m joined by our CEO, Yanick. Hello, Yanick.

Yanick Abraham: Alright, Rosalie! I’m very excited for this one. The key is back. And, just coming back from Dreamforce, where they kept saying, this is the most exciting Dreamforce so far, and I’ll borrow it, and I’ll say the same thing about this webinar. I think it’s our most exciting webinar so far, so very excited to be here.

Rosalie Chassat: All right, and we have… maybe we can go to the next slide. We also have Nathalie that’s in the chat with us today, which will be answering our questions, and I know many of you have questions about MCP, so please send them in the Q&A and the chat today. We’ll answer as many as we can live. I’ll bring them up to Yanick, and we’ll follow up after if we are not able to get to them. Also I want to mention, because I often get this question, the session is recorded, and we’ll send the recording afterwards, so you can share it with everybody else, and all the past webinars are on the website. I feel like this webinar, as well, similar to the one we did in June, is going to be one that people do watch a lot afterwards, right?

Yanick Abraham: I hope so.

Rosalie Chassat: Let’s start off with our first webinar. We want to know, let me just, launch it here. As always, we want to know, is this your first Klient webinar? Yes. No, I’m a regular. No, I’ve been to a couple, so just take a few seconds to answer that. Perfect. We do have a few, webinar first-timers, maybe it’s the topic that caught their interest from the webinar invite. So, welcome, if this is your first time, we do have a… yeah, quite a few, almost half of the people are first-time attendees, so, cool!

Yanick Abraham: I’m everyone.

Rosalie Chassat: Alright, so as you may have noticed, today is not a typical feature webinar like we do normally, so we are on version 51.11, but version 51.12 is not available yet, so we’ll cover that version on the October webinar. Therefore, today, we are focusing on something that we get a lot of questions and support and on customer calls. Which is, rolling out MCP across your organization. So we had very positive feedback on the last webinar, so we’re just doing part two now. Agenda, yes, sorry. We’ll start with a few takeaways from Dreamforce. Yanick was there, so he can tell us all about it. Then we talk about how we rolled out MCP at Klient, where agents fit in, lesson learned, etc. And then we’ll also, talk a little bit about customer community updates at the end, and then what’s next, a couple of teasers on, next version. Alright, and I think we have beautiful pictures for our Dreamforce little recap.

Yanick Abraham: Yeah, beautiful pictures. Not enough pictures. I figured out when I came back that I only have one picture with Raul or my colleague, so not enough. Still, those two pictures I took, they talk about… a lot about AIforce and Claudeforce. It was a very interesting Dreamforce, a lot of stuff happening, and we’ve been… I’ve been lucky enough to meet some of you guys, so some of our customers were there, some of our partners, we even threw happy hours on Monday, and it was nice meeting you, all of you. And the main key information here is that AIforce, the way I see it, is it’s kind of a way of saying, our user interface sucks. Go and build your own. So, this is pretty much what it says. They used to call kind of this kind of technology Headless, so being able to access data from everywhere, so, Claudeforce, Slack, Agentforce, it’s all a way of reaching your data, your Klient data, your Salesforce data, but without the standard UI that we are used to. So now the UI is being replaced by on-demand UI, by new experience UI, and we’ve kind of been doing this with you guys, right? The Snap workspace, or the workspace, or even the Planner, those are experienced, because we knew that Salesforce and UIs are lacking, so we provide new way of visualizing data, and we are pushing more in your ways as well. But the thing today is really talking. width. MCPs with outside agent. to Salesforce, through AIforce, and Claudeforce, it’s really the merger of the two, the two beasts, which, for us, it’s great. We love Claude, we love Salesforce, and they do make a nice play, and it’s a packaging thing, so when you Claudeforce, you can buy it, and it comes with… you can access Claude license-wise. through it, even though now we can still access it ourselves for stuff through Otherwise, but it’s a new packaging, a new way of generating revenue.

Rosalie Chassat: Yeah, I’m curious to know on the webinar today has attended, Dreamforce, or.

Yanick Abraham: I was curious.

Rosalie Chassat: happening, if you had any, like, takeaways as well that you’re kind of in line with what Yanick mentioned, or you took away some other things, we’re curious, like, feel free to just pop it in the chat. Always great to hear. Alright, I think that’s,

Yanick Abraham: Oh, yeah, if there’s no question, we’re gonna move to this one. This is the, Go ahead, Rosalie, and the next event will be in?

Rosalie Chassat: Yeah, we have it. Yeah, we have a slide for that. Alright, so as you may know, we attended a few of those Salesforce events throughout the year. Our next stop is Agentforce World Tour Amsterdam, in November, 26th of November. Yanick and I will be there, along with Raul and Krystel. We will have a booth. If you’re in Europe, or planning to be around, or you have someone from your team that’s in that area, have them come, meet us, and you can just pop it in the chat if you’re interested, or just reply to the recap email. You do need to register, it’s a free event, but we included the link, we’ll include it in the follow-up email, so…

Yanick Abraham: Most likely, we’re gonna be available as well for meetings on the Tuesday and Wednesday, so I’m probably doing a happy hour as well, so if you are around, please come.

Rosalie Chassat: Perfect. Alright, so let’s get into MCP. As I mentioned, this is a bit of a Part 2 webinar from the Part 1 in June. So, back in June, we ran a session called Connect Your AI to Klient PSA with MCP. We’ll drop the link somewhere in the chat, but it was included in the invitation. It’s a good recap if you missed it. What we covered there, just so everybody is on the same kind of level, what a MCP is, so we explained a bit of that. Then MCP versus agent. MCP is more flexible, lives under your AI tool, versus agents are more guided, live inside Salesforce, or now other places, like Slack. We also talked about security. This is also a topic that comes up a lot, so feel free to review, that webinar, and we have documentation on that. And then throughout the session, we did run a few polls, so I thought it would be interesting to just pop it on screen so people can see what you guys thought 3 months ago, and then we’ll do a couple more polls today, and maybe share that after the webinar. And then, yeah, we’re 3 months after, a lot has happened, so, yeah, picking up from there, I think I’ll just, drop a poll, or first poll, actually, to kick off. the next section. So we want to.

Yanick Abraham: Yeah, I’m curious to see, yeah, curious to see the result of this one.

Rosalie Chassat: MCP before today, and that’s the exact same question we asked 3 months ago, and I imagine the numbers would be higher for the answer yes. That’s my theory. And then the second one, who’s driving AI adoption at your company? Because, talking with customer, we do see that it’s very different from organization to organization. So just curious also if the person driving is the person on the webinar, or it’s some other thing, or it’s, like, a big teamwork effort. Yeah, so let’s see what, people are answering, and we still… okay, it’s a bit… maybe we can come back to that, those results at the end, I can pop them back. And I’ll leave it for…

Yanick Abraham: I have already the PSA slide, the stack open already.

Rosalie Chassat: Perfect. Alright, so let’s look at the PSA stack while people, keep answering that poll. So this is just… we get a lot of questions, again, to know, like, what’s available as of today when it comes to Klient and AI. So we have it all in one slide. So first of all, we have our AI agents. PLANNY1, our project management agent, TOUCHY1, and then the big news this month, finally, TIMEY1 is officially available. This is our timesheet, agent. So those three agents, available to purchase and start using in your organization today. Yay. Okay. Second, Klient, MCP and skills, which live in your AI tool, like Claude, Copilot, Cursor. The MCP data one is the one you want to get. It connects your AI to your Klient projects, tasks. time, assignments, cannot delete anything, and then we have those four skills that are currently available. So the project one, anything for project building, updating your projects, time for time logging, touch points, so preparing for your calls, following up after, creating tasks based on that, and then, the assignment one. Which is newer, for assigning people to projects, set their schedule, etc. Personally, I use the project skill, the time skill, and the touchpoint every day, pretty much. So… and then, I guess it’s different for each person. Yanick probably uses everything in there.

Yanick Abraham: No, you’re gonna see what I’m using, this is the demo I’m eager to show you.

Rosalie Chassat: Then the third one is the Knowledge MCP. We haven’t talked too much about it, but it helps you create and maintain all your Salesforce knowledge articles, so there’s the MCP and skill associated. And then on the roadmap, just so you know, these are the three things, so SCOPEY1, STAFFY1, and MCP Admin that are coming soon. Alright, if you want to know more, the fastest way to, like, get up to speed with any of these, if you’re interested, is just to go to, the Feature Hub, the community. Feature browser, and you have, like, one click for every single one of these, and then you can dive in and have fun. Alright, okay, let me close the… alright, let me close the poll, and I’ll share the results in the follow-up email, so that’s one more reason for everybody to read that. Alright! Let’s get started, so this is where I… hand it to you, Yanick, and you can show us all the things.

Yanick Abraham: And I’m real, I’m still the human in this segment. Today, the key thing is really to see how projects and how can you really articulate and deploy MCP to your business, but you will see that Klient project is really the core and the center of that AI deployment, at least here at Klient. And this is our… one of our taglines, where human leads, so when you’re gonna be able to take the decision, you’re gonna be able to approve stuff, so all that section here of decision logging, so the human leads and approve what the delivery and the AI is doing, it’s key, and the project put it as the core of this. So, it’s really the centralized hub where you’re gonna be, all your information About time, scope, your milestone, but also your actuals. And then all the to-dos of the AI, or the to-dos you want to give the AI that might happen at any time of the day, you just want to register this to be done at some point, the project is the perfect place to do that. On top, it’s going to provide you all that audit tracking to see, okay, what has been done, by who, to be able to really track what your MCP are going. So the proposed solution of how to work today is something that works very well for us. It’s really about efficiency and delivering. The other thing you see here is just a sidetrack of that efficiency. On top, it gives you other things, tracking, but it’s really about Delivering, delivering, delivering. So the project is core and what I’m gonna show you. And I’m happy because you brought it back, Rosalie, so the Klient inception, we’ve been talking about this. You might have one of the spinners that I have there on the back, so from the movie Inception. It’s a dream within a dream, so a lot of you guys are in consulting. We are… we’ve always… we’ve all been in consulting in our company, and we are doing some consulting, or in SaaS. A lot of people are in SaaS as well, and we are from We’re a SaaS business, so we are you in a sense, so that’s a dream within the dream, so we apply this, we are customer zero, and what we’ve been doing so far with AI, and that pretty much everything you see here, it’s all driven by project, and within those projects, we execute skills. And those skills, really what they do is that they manage processes. And the process that we are currently, I don’t know if it’s a real word, Agentified, so automated seems outdated, so it’s Agentified, so they’re running by agent, are the one you see here. So, from the idea, you register on our community, so our customer registering ideas on our community. From there, you’re getting scope by SCOPEY1, then it’s been refined by SCOPEY1 internally, then developed, QA’d, documented, so you all know that the documentation is generated automatically. The comments you put on the documentation is also going to be updated automatically. Plus, now we’re getting screenshot and close to video on the software dev side. In marketing, we control the entire website, including all the Google Ads, Bing, LinkedIn ads budget, so it’s all been driven, managed by objective in a project, that the AI check the objective in the project to make sure that we don’t go over, overspend, and things like that. We have a new AI sales agent who I can talk a long time. And maybe the webinar… the webinar one you can talk about, Rosalie?

Rosalie Chassat: Yeah, webinars, every time we do one of these, there’s a lot of things to do before and after. We just have a skill that basically supports doing that. Still have a bit of manual when it comes to the actual content creation, I think we still do that on our own, but everything before, all the publishing on the website, on YouTube, all that afterwards. We do leverage, like, that skill, and we have some associated projects, like, for LinkedIn content, for example, we have projects and Klient, so that works pretty well for me.

Yanick Abraham: And all this is based off a mix of processes that we’ve built together within the skills that our team now… I’ve changed my title at the beginning, I don’t know if you noticed, it was written Skill Maker, because I’ve been doing a lot of skills, creating a lot of skills, but now what’s great is that the skills have been delivered, and now Rosalie is updating skills, everybody in the company are updating skills, so we get better and better and better every day. And all those skills, they pretty much all use Klient Project, MCPs, the Klient data MCPs, the Klient project skills, plus other thing on top. So we are redefining all our department with it. And if you are interested to know more about this, if you are interested in doing documentation automatically, let me know. I would be happy to share with you what we’re doing, and if this is something that might fit your business as well. Okay, I want to set the table for today’s demo, so the use case is, it’s a project that we have internally, it’s an internal project. It’s the revamp, or the constant improvement of the feature hub, so the customer community that you guys logged in. So, this is an internal project, but it could easily be a buildable customer project, and for those of you who are a Salesforce partner, you do that every day. It’s a customer experience you build for your customer. So, there’s always new feature that’s wanted, it’s on the… now we’ve developed it, it’s live, but it’s… there’s always maintenance, there’s always new functionality, so this is the use case I’m going to present, how we manage those. What we do in terms of how we schedule, we use simple tasks and scheduling on phases. You’ll see very simple. The tasks are the to-dos, so the AI can really drive and be granular enough. So for those of you that don’t use Tasks, you will see that with AI, the use… the need of using tasks is there, and is absolutely needed. And then, task assignment, you won’t really see it during the demo, but task assignments are swapped between agents and human, so this way we know who’s in charge of that task at which time, because human and agents are working together to deliver a project. It’s not fully automated, we are always in control, we’re always dealing. The doing is what the AI agent do. So, this is one example, but we have other projects, like the one we have just mentioned. The sales agent is driven by one of those A project, documentation, website prevent KLARA, the support agent, and even the Ideas Exchange. Works in that kind of formula for software development. And soon, I mean, this one is a… We call it onboarding project. Internally, you have a tab of some of you, onboarding project, but it’s kind of an improved project, the code name is not defined, but all this, with our Agent Guide ID that’s gonna be coming later in maybe 2027, is going to be really something that’s gonna help you out onboard, so this is… this is also kind of something linked to it.

Rosalie Chassat: By the way, if you have any questions as Yanick is talking about all these topics, just pop them in the chat.

Yanick Abraham: Okay, so this was the frame of what we’re going to be seeing in the next, the demo, so really a feature hub improvement, so a customer community improvement, so kind of a software development project. Billable or not. And what I’m gonna be using, this is what kind of setup we have for everything. So, we are running on Claude Code with VS Code. But we all use Project, and they, for us, we have a naming convention for those kind of agent projects, they start with A. We also have an associated Slack channel, so for team people, Microsoft Team, you don’t have this yet, but eventually, but we have a Slack channel that we can link, I’m gonna show it today. We have an awareness channel, so when we develop something new, let’s say we have a new adoption dashboard for your usage of the product now, so we’ve published that internally so everybody’s aware, so this we automate that as well. And we have a notification channel, so every time that you guys post a comment, for example, or a new idea, we receive a ding in Slack about this, and Rosalie loved that for that dopamine it gives. Every time you guys write, so keep writing so we see what you’re doing. So this is kind of the setup for developing with AI that we find very useful on our side. Then, for this demo, we are going to be seeing the MCP Data 1 and Knowledge1 for documentation. It’s always pretty much a tool I’m always using. Every time that I want to refer to the product, it looks at the Knowledge articles and all the L particles. The skills: Project, you will see the demo of it touchpoint, I won’t demo it, but sometime when we have calls together, we’re gonna record, get the transcript, and this is gonna be the input for the task we’re gonna create. I’m gonna show you that. Distribution, it’s on GitHub now, on our side. We’re all a Claude shop on VS Code, and we’re, like, running our stuff local, so we don’t have any skill hosted yet, besides that GitHub. So this is where sometimes you might have, We might need to do some work with you to define your specific setup, so we’re happy to help you out if you want to do something remote. We have Copilot that is available, but we don’t run Copilot internally. Okay, so that was a long intro.

Rosalie Chassat: That’s the ingredients for the recipe, and now we’re gonna…

Yanick Abraham: Yeah, exactly, so at least you have the frame. So now what we do. So this is, seriously, this is my real life, and I really enjoyed it. I’m a technical person. First, I guess, at least it shows now with all that AI stuff. I love doing a lot of that work, and that project, FeatureHub, it’s kind of… it was one of my pet projects, my side project, so now I just maintain it. So I’m gonna show you how I’m doing it on a daily basis, and what I’m… the way I’m running and delivering feature as a producer, because we don’t call our people developer anymore, we call them producer. I’m gonna show you how I deliver this project. Okay, so I’m back to that beautiful UI of Salesforce. We understand now that why they want to replace it and give us better tools. So this is a standard Salesforce record, there’s nothing there. But let’s jump in into that feature hub project and look at the resource plan. So that’s my project that you and I, Rosalie, have been working on since a few months already. And if I go back in time… You’re gonna see that it’s a basic project in terms of scheduling. We have that information here, that we have that display density here, that, That’s a new feature that I want to plug in, that’s very nice, having that menu here. And you can see that Rosalie and I, we have a few tasks, so the initial design, development, and ongoing development. So those are human tasks, they still remain in those kind of AI projects, you still need human tasks. But very simple, and you can see that I had 6 hours per week, approximately, associated to that project, so I know I can… I need to block my time for 6 hours per week to develop it. And let’s now take a look at the same project, but with the workspace lens. So you can see here at the top, we have that dose scheduling we just saw, so with initial design, development, ongoing improvement, that’s all assigned to you and me. And then we have a typical project with the feature, the functionality, or the thing that you’re delivering. So this, you guys all have this, or sections, phases, or whatever of things that need to be done. But what’s… and you can see there’s a lot of unparented tasks here, right? So… This is what we’re going to be doing with the AI, we’re going to be a little bit of cleanup in this, this UI. For those kind of AI projects, I don’t use a standard, my standard scheduling preset. I would use one avenue, a preset I used for the A project, because the column it’s important to work with, with the AI are a bit different. You don’t care about duration, you don’t care about start date, end date, because all… pretty much everything can be done within, what. Two minutes and a half, so it’s… it gives another way of looking at things and doing things, so it’s really more… working more with statuses. So it’s really looking at your task of, okay, what needs to be done for that kind of project, because the delivery of that project can be done by the AI. That’s a big differentiator of this kind of project, but let’s say we have for that e-learning, we want to add a sharing link, I would just need to move this task to ready for dev. And by moving it ready for dev, it means that the machine can pick it up. So the human is still in control of the statuses, but this is how I work with my agents, and how I work on, okay, what we should be doing next. I select a few from here, and then I’m gonna switch to the MCP in a second. So you can see the next step, the description, the proposed solution, and the statuses. Priority are all important columns when you’re working with AI. Okay, so, I think I can switch… Okay, any questions so far in Rosalie with Encore yet? I didn’t get into the core yet. Okay, so now we are in… Claude. Okay, Claude Code in this case, but this can work with any of your AI, it’s all the same thing. And I’m gonna launch two things right away. So, first thing I’m gonna show, it’s my MCP server, so I’m gonna show you how many we have. So we have a lot. The one we are talking about today, and the one that you would be using are the MCP Data 1, in our case, Knowledge One as well. And you have, and then we have a bunch of other MCP that we’ve built for all the other skills that we’ve shown you before. And now what I’m gonna do is I’m gonna talk to Claude, and I’m gonna ask him… let me zoom in a little bit… So, I’m calling the skills, Project 1, and I’m saying organize the A feature project, so every task as a parent, use existing parent, or create new ones. Show me the plan first. Okay, so this is a problem that all of you have, right? When you don’t use tasks because it becomes messy, or we use tasks and then they become outdated, they’re not followed, they’re not… so this is the exact use case of calling this prompt here. While we’re doing this, and it’s gonna take a few seconds, it might go faster than I thought, but I’m gonna launch another one here on the side. This one is to… I’ve talked about Slack, so read the Slack channel, a feature of, and propose new tasks to add to the project. Okay, so this is… now it’s doing its job of fetching the communication channel, and while it’s fetching it, we’re gonna look at that channel. Okay, so this is a channel that Rosalie is involved a lot, because she has a lot of ideas, which is great. This is the new AI, it’s the ages of idea and creation, so we want new ideas, because we can implement them. So you can see here, there’s a few, of those ideas that have been registered in that, that Slack channel. But actually, I don’t want to work this way. Let me go back to the project, the Salesforce project. You can see here on the right side, we have a Slack channel embedded. So that’s one of the cool things as well with Salesforce. If you have Slack, you can enable channel to be on Slack. And when you do that, they’re gonna… the channel are gonna be named the exact same name as the object you’re in, in our case, this project. So that’s why that project is named A- Feature Hub with caps, letters, because it’s a project link between Slack and Salesforce. Okay, so this is already very cool. Let’s come back to our prompt, there’s one that I finished. That’s the idea of also working with AI, so you always have multiple windows running. Sometimes I can have as much as 12, 14, some of those threads are all running in parallel, because they all take times longer or shorter, but this is how I work. Okay, so you can see, in that triage, or that organization, it recommended 3 new parent tasks. One for the store, one for the new gamification and adoption that Rosalie is talking about, and some for doc skills and tooling. Plus, it proposed to move some of those tasks in those, bigger buckets. And then it just asked me, It’s always a different answer, so…

Rosalie Chassat: placement I’m least sure about.

Yanick Abraham: I’m sorry?

Rosalie Chassat: the section where it’s, like, the ones you’re not sure, those are the ones I typically, when I get the response, those are the ones I look first, the one that’s not.

Yanick Abraham: Yeah, exactly, those here. But, let’s say I use enough of that system that I trust them enough, I’m gonna just say go, because this is what it tells me to do, and then it’s gonna start doing that cleanup, okay? And actually, I should have said, just do that list, but, I’m pretty sure it’s gonna do it properly. Okay, the other one is very interesting. Okay, so the other one, it took something messy, which is Slack post, and they could be very messy Slack posts, and it took… and there’s also a screenshot in it, right? But it took it, and then it’s telling me that, okay, based on that, we should create 3 tasks. One for adoption Dashboard, actually two for adoption Dashboard, one for the idea section, the other one for the case list, and one for gamification. So, my response to this… V… Give me a sec. This is the power of it. So, yes, create the task. But don’t parent them, because I don’t want to… I wanna find them, because the other one is doing this cleanup, so I don’t want to get overwritten or re-parented. And then, also put AI’emoji on our Slack, so here on the right side, to show that we are working on it. And the rest of the port, start fixing the easiest one and change status to ready for QA when done. And put a checkmark emoji on the repost. So this is exactly how I work. Usually, I don’t type all of this, because it knows my way of doing things. So you see down the eyes emoji are starting to show, bing, bing, boom, so it’s been… created those tasks, you’re gonna pick the easiest one, and it’s gonna develop it. Well, the one on the left is updating all my tasks, you can see he’s doing that update, he’s using MCP Data 1, you can definitely see what he’s doing. And then, if I’m going on my featured project. I’ll be, Yep, good place. I’ll be starting to see some big changes in the project. Let me just refresh it. So, all those, non-parented tasks, we had a lot, so they’re all getting… being rerouted internally. We’re gonna give it… it’s a big job, so I’m assuming it’s gonna be done within 10-15 minutes. 5-10 minutes, but it’s starting doing that, reparenting of tasks and that classification for you. Exciting?

Rosalie Chassat: Yeah, do you want to take a question? Is that good timing?

Yanick Abraham: Yes, please.

Rosalie Chassat: We get a question from Bob. How do you… and we get that question, I think, from every single customer, how do you manage or predict token usage with the MCP connection?

Yanick Abraham: My answer, or,

Rosalie Chassat: Your answer… I know what your answer is, but yeah, your answer, and then maybe what we recommend for.

Yanick Abraham: Seriously, everything we do, we are on the, I’ll be transparent, we are on the Claude Max package, I don’t know, 200-something per user per month. Everybody of us are all on that package. And I’m the only one busting my limits in terms of tokens, so it’s… you can do whatever you want with it, we don’t want. At this point, we don’t care about Claude token usage. Because it fits within the budget associated with the pricing that we pay for, so until we go over, then we’re gonna start, like, defining. When it’s meow busting, I’m… I’m allowed a special budget, so I’m fine, but so far, so good. Moving forward, though, and this is something I don’t want to necessarily talk about today, because it’s going to be in the future release, but the same way that you guys, human, log time on task. We’re gonna ask agent to log token and time on task, so the agent will be required to do timesheet, because it’s… we’re gonna start seeing it, that it’s not free, it’s not only $200 per month, it’s more, so we want to affect those costs into the project. So you can recharge your customer, or, so you can, know at least your costs, so you can really calculate your margin. So this is how we’re gonna end… be doing. At this point, we don’t care internally as Klient. Moving forward, we do care, and it’s going to be done through a agent logging their time and token.

Rosalie Chassat: Yeah, and I think for customers, because we get that question, if they’re curious, you just run the scale, or you run whatever you would want, and then you just ask Claude how many tokens I’ll calculate, and you can ask it to calculate based on your use case, and then you’ll get a pretty… good answer, but it’s hard for us to tell you exactly because, again, like, we’re running a bunch of these different ones, so each person does things differently. But yeah, you can ask Claude.

Yanick Abraham: I wasn’t sure how fast it would go, but it already completed the one, the first one, so, I asked him to do another one. So, you have a new feature now coming in, and there’s a way to track it in the system, so let me just look at the name of the task he created. I can just ask him, Giving me… URL of the task. And seriously, the way I talk, it’s really minimal in terms of information, and usually you understand me very well all the time, so… and the… I’m not sure how far you go with using skills in those systems, and then Copilot might be different. Sometimes you do want to invoke the skills explicitly to make sure that it’s going to do the thing. When we’re talking about project, it pretty much defaults all the time to our skills, so you don’t need to always invoke it. But if you want to make sure, It’s better you invoke it, depending on what you want it. I think, well, it’s working, maybe there’s something else we can show you, that is not MCP.

Rosalie Chassat: Something I do want to mention, I do some… I know some people on the call are not using Slack. In this example that Yanick is showing, it’s pulling the information from Slack, but you could also pull the information from last 7 days of call transcripts and do the same process. Like, your source can be whatever. In this case, we do it for Slack, but it could be any other source. That you have access to, so…

Yanick Abraham: Yeah, exactly. So, for now, it’s exactly what you said. It’s a communication method. It could be look at my email instead of looking at my Slack, and Teams is just that I’m not connected to Teams, but Copilot is connected to Teams, so if you do the same thing with Teams, it’s gonna… it’s just a channel of incoming for creation of tasks, huh? Just because, why do you want to create tasks? Because you don’t want to have, like, task in a team, task in your email, task in Klient, task in… Just create tasks all the time in Klient, and then that’s problem solved. And you can see those three tasks are the last one created, so the adoption dashboard, so your case list opens case first, then by date, ready for QA, Rosalie, so if you just want to go see, it is working, so… This is how we develop now. So, kind of an autonomous project delivery, or AI delivery, not sure how you want to frame it, but this is what happened. Okay, how do you use an agent with this? So if we go back to that channel here, Slack channel, I’m doing it here, again, I could be directly in Salesforce, same thing, but what I’m gonna do is I’m gonna invite PLANNY1 to that channel. I’m sorry?

Rosalie Chassat: No, we couldn’t see your Slack.

Yanick Abraham: Oh, you see it now, or…

Rosalie Chassat: Yep, yep.

Yanick Abraham: Okay, so you’re on the right. So, I do want to add PLANNY1 to that conversation. Yes, please? I want to talk to PLANNY1 about if my project is going on track or not. So now, this is it. I could have asked Claude, but sometime I’m here, I’m on my machine, I’m not on VS Code, I’m on my mobile. I’m going to talk to Slack. So that’s the thing with agents and skills and code, so… you can use diff… you can use both in different moments, depending on what you’re doing. So, in this case, let’s say I’m on mobile, or I just have Slack open, and I just want to know. So, let me just view my… We’re a year. Very good. So, just a simple request, but how many tasks are open on this project? So, PLANNY1 is going to work, and then it’s querying all your data, and it’s going to bring back that, that reply. As you can see on read, I think this is new, the red one. Sharing trend… Hello! No. What happened?

Rosalie Chassat: Because you didn’t tell him… yeah, he couldn’t help, because he wasn’t on the channel yet, maybe.

Yanick Abraham: Once, very well during my, my prep. It’s always like this one. Finishing with the agent.

Rosalie Chassat: And you reply.

Yanick Abraham: I’m sure Sherish is looking, saying, what? Oh, you forgot this? Okay, so, let’s see if it reverts, and let’s, but that’s, are you… Oh, there you go. I’ve never seen that share intro. So, my, previous, report was better, so there are 94 tasks open in the future project, which is true, but the idea with PLANNY1 is that within Slack, within conversation. You can ask him more information, so, I could ask him, show me the overdue task or things like that, but this would be, another demo for agents. Today, with a focus was really talking about MCPs, and how you can automate stuff from, from your LLMs directly with Klient. Anything I forgot to mention, Rosalie?

Rosalie Chassat: I think, yeah, just mentioning, disclaimer, like, this is one way of doing things. We have, like, a billion other projects, and we are conscious, like, each person will adjust, but hopefully from this demo, there’s, like, some things that you pick up that you can apply. Maybe not everything, but, like, that’s the purpose of, like, showing you examples, And such. And then, this is…

Yanick Abraham: That’s really human and agents working together on projects, and while it’s not perfect, it’s, like, evolving every week, we’re getting better every week. Now I have to launch those things manually, and then next release, you’re gonna see we’re gonna be able to schedule those jobs automatically. So, same for PLANNY1, he’s gonna… he’s waking you up at 8 AM with his message, but our other Claude Code will be, doing, the same kind of routine on a scheduled basis with sales, Klient schedules. Okay, that’s it for the demo.

Rosalie Chassat: Let’s go to the next slide.

Yanick Abraham: You wanna take it, or you want me to do it?

Rosalie Chassat: How to get started? Oh, I don’t see the next slide. Oh, there you go. I think there’s a delay.

Yanick Abraham: There’s a building.

Rosalie Chassat: Alright, so, yeah, how to get started. We mentioned this in a couple of other webinars, but just a couple of things. First thing, so after today, you’re like, I want, I want in, let’s do it. First thing to do, pick your team. So, this is our recommendations, right? So, choose one or two reliable people to champion the rollout. Maybe you guys are on the room right now, which is great. And then, this way you can get some early wins before, you open it up to everybody else. Then the second step would be to get the MCP and skills, which you can do from Feature Hub, My Product, and then you just select, download, you buy it, and then you just follow the install guide. If you have any questions, we’re here to help. Third step, create your first AI project in Klient, obviously using, like, the skill to create it, don’t do it manually. Just like Yanick showed you, well, we showed the adding tasks, but creating a project is kind of the same principle. And then that’s where your team’s AI work will live, and where you’ll see everything that was done. And then the last step is executing the work. So right from your AI tool, like I showed you, whether it’s Claude, Copilot, Gemini, and that’s kind of the rollout of it. If we had to do, like, a recipe for it. And then one thing as well, like, Yanick mentioned, like, you have this champion, but then eventually we encourage you to, like, share those skills, and then make people accountable, and then each person become responsible for their own skill. Alright, I think we’ll put in the chat the My Products, in case you guys still don’t know where that is. It’s from the community, if you want to buy any of these. And then I wanted to do a few updates on the community, if we have the next slide. Maybe there’s a delay there, but I’ll just.

Yanick Abraham: I wanted to see how all my cleanup was going, so you see the cleanup is done now.

Rosalie Chassat: The project that we were just showing, these are all things that we keep updating on the Feature Hub, so obviously there are things that were updated, we’re not communicating with you. So a few things as a reminder… do we have the slides? Yeah, I think I briefly touched on it, but our Ideas Exchange, this update, we wanted to tell you, so in case you’re wondering what’s happening there, there are so far 129 ideas that were submitted since, that’s live, so it’s March. That’s quite a lot. 46 of those are already in development or live, and this is just proving to show, like, please do continue submitting your things, because we will develop them. And then we have 14 that are from the last 30 votes, and of course, people keep voting for them, so they go up, etc. Then, let me just go back… oh, another new thing, is this whole point, concept of points and badges. So we now have that. You can access it from my profile when you go in the community, and you can see for you and the rest of your team, so that’s pretty cool. We’ll add more to that. And if you have feedback, we’ll create a task in the Feature Hub project and incorporate it.

Yanick Abraham: Wow.

Rosalie Chassat: The e-learning as well, so you know by now, we added some e-learnings, but we now have, like, a leaderboard that you can follow and see, like, who’s doing the e-learning, etc. And self-registration, I’m mentioning it because we do still have a lot of requests and support, people asking us, hey, I want access to the community, you just need to go to the login page and create your own community login. Or, so anyone that needs access, that’s the way to do it. And then the other thing I wanted to do, I guess we’re on the next page for what’s coming next. Are you there yet?

Yanick Abraham: about the language change?

Rosalie Chassat: Yeah, okay. I just wanted to do a little teaser, because, again, we didn’t cover 51.12 today, because it’s not yet available. We’ll cover it in October. But until then… Yeah, 51.12 coming October 9. A few things that are coming that are pretty exciting, I handpicked them, think they’re pretty cool. Draft Mode in Resource Planner is coming, so you can prepare and, like, set up your resource assignment, and then you can do a couple of those, see what it looks like, and hit the publish if you want, or just don’t save it, so kind of scenario-based type of thing. Predecessor column, we got a lot of, question for this one as well, so having a column where you can kind of key.

Yanick Abraham: Yes. When the…

Rosalie Chassat: Dependencies instead of, like, having to drag it, which is the old way, I guess, so that’s coming. Group by billable and My Timesheets. We’ve added a lot of different ways to sort and group things in my timesheets, so this is one that we got a request for. Presets and staff workspace, it’s coming. Pretty straightforward, that one. We love presets, as you can tell, maybe. And the last one I added, Survey Insights. This is a screenshot of what it looks like. This is coming, a release after that. But it’s, essentially, like, surveys report are a bit tricky in Klient, and now we’ll have a new interface for consuming all that data. So if this was a stop to you using your surveys, when that’s coming, maybe that’s, like, kind of the little kick you needed to start using your surveys. But this is an example of what it looks like, so pretty exciting. That one I really like.

Yanick Abraham: Is it using AIforce and Claudeforce, that, Rosalie?

Rosalie Chassat: Well, we built it, but no, no, I don’t think it is.

Yanick Abraham: Same concept, right? New UI, new way, better way, easiest way, more…

Rosalie Chassat: Well, this is something we wanted to do for a while, and we couldn’t, but now with AR, it’s much faster to get to the end. Yeah. Okay, this brings us to the Q&A. I feel like there, like, we didn’t get questions, which I’m really surprised about, so maybe people are, like, reflecting on what we just showed. That’s fine, just a couple of things, like, I know people ask that maybe I just shout is, people often ask, like, can your AI delete things, or, like. It cannot, so when you’re using the MCP, it cannot delete tasks. our projects.

Yanick Abraham: Our MCP, right? Our MCP.

Rosalie Chassat: Yeah, yeah, yeah. So sometimes, if you ask to update a project, it will, like, add new tasks and say, but you need to delete those ones, because these are, like, the replacements, so it cannot delete. Also, a question we get asked, like, can I build my own skills? Of course you can. Usually what we recommend is, like. Get the skill, but then we encourage you to, like, improve and, like, save it as a new skill.

Yanick Abraham: Always. All the time, that’s the key.

Rosalie Chassat: there’s always things that are specific to your org, or even specific per role, so I’m using the time skill, but I’ve customized it to my specific role. Then, yeah, I think that’s… Yeah, there’s a bunch.

Yanick Abraham: It’s been around enough time anyway.

Rosalie Chassat: Yeah, if people have questions, please send them. And then, wrapping up, I just wanted to… role that, we have two questions, so I want to start… well, let’s just do the webinar feedback. In the webinar feedback, I know there wasn’t a lot of questions, so I’m curious to know, please take a few seconds to respond to the webinar. Is it because, super interesting and overwhelmed, and you didn’t want to ask questions, or, like, we were completely off track, and this was not useful at all? Like, whatever it is, please… Respond to the feedback, so we know how this works.

Yanick Abraham: I won’t take it personal. We want to improve.

Rosalie Chassat: So we’ll leave you a couple of seconds. And when you get to your MCP agents rollout, please, like, share your insights, lesson learned, questions. We love to hear about what you’re doing, so don’t hesitate to, reach out on that end. I will also encourage you to register to the October webinar, so that’s in about 4 weeks. So you just follow the normal link, we’ll send a recap email afterwards, that included anyways, but hopefully I see you in, October. Yanick, did you have anything else you wanted to add?

Yanick Abraham: Oh, go MCP, go. It’s a lot of fun. It’s very powerful. Please use it. You’re gonna see it’s gonna change your life.

Rosalie Chassat: Alright, have a great rest of your day!

Yanick Abraham: Thank you, everyone! Thank you soon.

Free download

Download the webinar deck

Get the full slide deck from this session — we'll email it straight to your inbox.

PDF · free · no spam

RC
Rosalie Chassat
Rosalie Chassat leads Customer Success at Klient and co-hosts the monthly Klient PSA Product Update Webinar. She helps consulting and SaaS services teams get the most out of Klient PSA — the 100% Salesforce-native PSA with AI agents.
Connect on LinkedIn
About Klient PSA: Klient runs your entire services business inside the Salesforce you already own — projects, resourcing, time, and billing on one platform, with live margin on every engagement and no data to sync. Explore the Salesforce PSA built for professional services.

See Klient PSA in action.

Book a 30-minute demo. Go live in 3 weeks.

$39/user/mo 3 wks go-live 100% Salesforce native