Happy Friday!
I had a reality check in a conversation this week with the leader of a professional services team. I thought my readers might also benefit from this reality check (or laugh that I’ve been so far removed!)
Most people I talk to are what you might call "AI-pilled": they're mostly tech people, builders at heart.
I always want to swap use cases and best practices with these folks. It helps me use AI more effectively myself, it helps me understand what’s going on in the market, and usually I have something helpful to share back as well. (If you want to have one of these conversations, my cal is always open 🙂 )
This week, I had one of these chats with a leader at a large SaaS company. She was the only person in her 50-person professional services org with a Claude license, and she wanted to make sure she was making the most of it.
Let me repeat that: the only person with a Claude license! Everyone else is still doing work by hand!!
It gets worse: she is managing multiple teams - and has a day job to do - in addition to trying to maximize her value from Claude. And her team got cut by 70% earlier this year because “AI” … but no one is using AI.
There were so many good AI use cases, too: initial product setup and config, better initial discovery, drafting initial case replies, creating first draft project plans, etc etc.
The blockers to more AI usage within the company were governance, control, but mostly (in my read) inertia. The broader company had an AI task force, but proserv is one small part of that company (and never first priority). Cultural hesitation around change is old hat to experienced SIs and proserv teams, and the AI wave is no different.
This was my reminder: enterprises are just slower. Most people - even most people in tech! - are in a lived reality that is so far from agent swarms that the present feels almost entirely disconnected.
If you’re doing implementations - are you seeing this too?
One of my favorite boomer-coded activities is pasting a screenshot from social media. You are all witnessing: I am entering my boomer era.
What We Shipped This Week & What's Coming Next
This week:
Supersession - For your super discovery sessions. Just kidding - for all of them!
We all know that discovery isn't linear. You learn things from one stakeholder that are later changed or are corrected by another. Previously, Glossa flagged situations like these as potential contradictions. But that’s not quite right - often, you make a decision live within the meeting and move on.
Now, Glossa better understands the timeline of your discovery process - using your meeting dates to understand decisions replaced by later decisions, requirements whose reasons have since been deferred or contradicted, and risks mitigated (or exacerbated) by new information.
When a requirement (or decision, risk, constraint, or finding) is out of date, Glossa now automatically marks it as Suggested Cancel, and tells you:
why it’s out of date
the up-to-date requirement that’s replacing it
For now, we want humans to review and give us feedback on supersession. If you like it, we’ll make it totally automated - so try it out and let me know!
Export requirements and categories - You know what’s annoying? Vlookups. Previously, if you wanted to get a list of all of your user stories with assumptions, complexity drivers, and details from their categories and capabilities, you had to export two csvs and combine them. Gross! Now, we have one button on the dashboard that will export a single sheet with user stories and categories, together. Thank you Charity for this idea!
A few quality fixes - meeting action items no longer get pulled in as "decisions,” there’s less overlap between outcomes and decisions, and merged decisions and findings are cleaner. If you’re seeing things that need tidying, let me know!
Coming next:
I want your feedback. Are you seeing too many contradictions per project? Do you want a more robust risk register? Do you want requirements tagged with the person who said the requirement, so you can filter on just things that that person said?
And: we have many different requirement counts: all requirements, v active requirements, v decisions/risks/constraints/findings. There’s a bunch of numbers across the UI - we are going to do a better job of making it clear what means what. (Thank you Aske for this idea!)
Weekend Reading
I’ve been talking to a lot of people working on AI-first Salesforce businesses. I wanted to try to map them - what am I missing? Would a business model map be more interesting?
As always, down to chat about all things services, AI, Salesforce, puppies, etc.
Have a great weekend!
- Ali