so glad you liked it - thanks for letting me know - always slightly hesitant to share very technical topics here (not sure if people are really into them) - but so glad this one resonated. Thanks a million for your ongoing support and friendship <3 have a great weekend
Loved this piece, especially the framing of context engineering as the difference between “expensive autocomplete” and an actual institutional brain for GTM. The CLAUDE.md + skills + MCP + hooks stack is such a clear mental model for how to get out of prompt ping-pong and into compounding systems.
One layer I keep seeing teams miss, even when they nail all of this, is the customer’s *own* language and motivations. There’s a ton of rigor around internal docs, ICP definitions, and GTM playbooks, and almost no time spent actually talking to customers, running Jobs to Be Done-style interviews, and encoding the functional, social, and emotional reasons they hire the product into that context system.
From my vantage point, the biggest unlock is when CLAUDE.md and the surrounding knowledge base are fed by real conversations: transcripts, call notes, and structured JTBD insights that clarify what customers are trying to accomplish, how they describe it, and what “success” feels like to them. Otherwise, we risk building beautifully engineered AI context around assumptions, not lived customer reality.
This sure beat all the "bumf" that people keep sharing on LinkedIn that keeps me wondering if this is all theory & AI generated or if someone has rolled up their sleeves and actually implemented a working solution. Maybe I am too old school and pragmatic. Start with understanding the current process, map it out, apply common sense, see if there are any quick wins, get buy-in from key staff, implement the core storage (CRM) make it the one version of the truth. Then start automating parts of the sales process that will make the biggest difference to the sales pipeline. Focus on simplifying and innovating as you go along. Monitor progress and celebrate success. Visualise the timeline & share it openly using Miro. Calculate ROI. Then Scale if stable.
Maja, you are amazing! 🙌 I've spent the last few weeks reading about this, and this is honestly the first article that actually made it click.
Thank you for being so generous with what you share and for always being one step ahead, helping the rest of us level up too 🙏
so glad you liked it - thanks for letting me know - always slightly hesitant to share very technical topics here (not sure if people are really into them) - but so glad this one resonated. Thanks a million for your ongoing support and friendship <3 have a great weekend
Loved this piece, especially the framing of context engineering as the difference between “expensive autocomplete” and an actual institutional brain for GTM. The CLAUDE.md + skills + MCP + hooks stack is such a clear mental model for how to get out of prompt ping-pong and into compounding systems.
One layer I keep seeing teams miss, even when they nail all of this, is the customer’s *own* language and motivations. There’s a ton of rigor around internal docs, ICP definitions, and GTM playbooks, and almost no time spent actually talking to customers, running Jobs to Be Done-style interviews, and encoding the functional, social, and emotional reasons they hire the product into that context system.
From my vantage point, the biggest unlock is when CLAUDE.md and the surrounding knowledge base are fed by real conversations: transcripts, call notes, and structured JTBD insights that clarify what customers are trying to accomplish, how they describe it, and what “success” feels like to them. Otherwise, we risk building beautifully engineered AI context around assumptions, not lived customer reality.
This sure beat all the "bumf" that people keep sharing on LinkedIn that keeps me wondering if this is all theory & AI generated or if someone has rolled up their sleeves and actually implemented a working solution. Maybe I am too old school and pragmatic. Start with understanding the current process, map it out, apply common sense, see if there are any quick wins, get buy-in from key staff, implement the core storage (CRM) make it the one version of the truth. Then start automating parts of the sales process that will make the biggest difference to the sales pipeline. Focus on simplifying and innovating as you go along. Monitor progress and celebrate success. Visualise the timeline & share it openly using Miro. Calculate ROI. Then Scale if stable.