Been using the Claude Excel plugin for a week and I genuinely didn’t expect it to hit this hard
Why I picked this
Victor's question lands hard: what happens when the augmentation becomes full automation? This practitioner just collapsed five days of financial modeling work into one day using Claude's Excel plugin — not by doing the work for them, but by becoming a debugging partner that actually understands circular references and cross-sheet dependencies. The productivity gain is real and measured. But here's the uncomfortable part: they're still constructing the sheets, still making the strategic choices about what to model. The tool is a spectacular co-pilot. The next step Victor's pointing at — where Claude builds the entire model from a prompt — feels both inevitable and destabilizing. We're watching the last mile of human-in-the-loop knowledge work, and it's compressing faster than anyone expected. The practitioner didn't expect it to 'hit this hard.' Neither did most of us.
Three lenses
I'd be prototyping the agentic version this weekend — prompt-to-model pipeline with validation loops. The hard part isn't generation, it's knowing when the model is actually correct, which is exactly what this practitioner still provides.
Five days to one day is a 400% productivity gain on complex deliverables. If my analysts can produce models this fast without quality degradation, I'm rethinking headcount plans and what 'senior analyst' even means in 18 months.
Everyone's celebrating the speed gain. Nobody's talking about the new failure mode: analysts who can't build models from scratch anymore, completely dependent on a tool they don't understand when it hallucinates a formula.
Companies
Key metrics
- •5 days reduced to 1 day
Why this matters for operators: Operators need to map which knowledge work tasks are in this augmentation zone now versus full automation in 12 months — the transition window is shorter than planning cycles.
I cover AI×GTM intelligence like this every Wednesday.
Get STEEPWORKS WeeklyMore picks
Humans will keep inventing new reasons why we must stay in the loop with agents
- Human resistance to full AI autonomy is not purely technical—it's psychological and organizational; companies will rationalize keeping humans in decision loops even when agents are capable
- The 'human-in-the-loop' requirement may become a self-perpetuating narrative rather than a genuine necessity, driven by organizational risk aversion and change resistance
- Product leaders at scale (Notion) are observing this pattern, suggesting it's a widespread phenomenon across enterprise AI adoption, not isolated to specific use cases
Comp plans for consumption pricing
- Consumption pricing fundamentally breaks traditional SaaS comp models—requires rethinking sales incentive structures around usage vs. contract value
- Four distinct contract structures exist (pay-as-you-go, uncommitted, committed, hybrid), each requiring different compensation mechanics and sales behaviors
- Enterprise consumption-based deals create tension: customers want flexibility, sales teams need predictability for quota attainment—comp design must bridge this gap
3 revenue motions your AI is only half wired into
- Model parity has arrived: OpenAI/Claude now trade evenly on core tasks, making 'better AI' a non-differentiator—the edge shifts to integration depth into existing revenue motions
- Waste is quantified: teams paying $17K-$37K/month for AI seats that never touch pipeline generation; real cost is opportunity cost of unused capacity, not subscription fees
- Lean teams have a structural advantage: cannot out-buy larger competitors on model access, but can out-embed them by wiring AI 1 revenue motion deep (pipeline → content → deals) with proprietary deal context competitors haven't seen
This analysis was produced using the STEEPWORKS system — the same agents, skills, and knowledge architecture available in the GrowthOS package.