Where the hours and the money are going, what to automate first, and what it would be worth to do so.
Across 14 people, Hartwell is burning an estimated 340 hours per month on work that is structurally repetitive — the kind of work where the shape of the output is the same every week, only the inputs change. At blended rates that is ~$34,200 per month.
Of that, roughly two thirds can be offloaded to a handful of small, purpose-built agents. Not a platform. Not a replatforming. Three to five workflows, built in order, each one ending a specific bleed.
The goal of this document is to name exactly which bleeds, what each one costs, and which three we would stop first.
Note — Hours estimated from a 45-minute discovery call with Sarah H. (Ops) and James W. (Research Director). Blended rate assumes a mixed team of senior and junior contributors. Conservative.
Currently pieced together from 11 sources by two analysts. Structurally identical week over week.
Senior staff rewriting the same 12 sections with client-specific variables. Highest-cost drain.
Turning 30–60 page source decks into 2-page client briefs. Pattern is consistent.
Weekly recurring slides populated from project tracker data. Pure data-to-template.
Transcripts already exist. Extraction + assignment is mechanical. Fastest possible win.
Transcripts auto-pulled from Zoom. Structured action items extracted and assigned into Linear with owners and due dates. Zero human touch after the call ends.
Scheduled crawl of 11 tracked sources. Claude drafts the weekly brief in your existing template. An analyst spends 20 minutes on final review instead of 12 hours writing.
A senior fills in a 9-field brief. The agent drafts the full proposal in your voice, pulling from 180 past proposals and case studies. Senior edits instead of writes.
Note — ROI scores weight hours saved against cost to build and time to production. A 10 is an obvious build. Below 5, we would leave it alone this year.
If Hartwell were our client, we would build Q1 (Meeting notes → action items)first. Not because it saves the most hours — it doesn’t — but because it will be running in production in three days and every person on the team will feel it the next morning.
That matters. The thing most AI programs lack isn’t a roadmap. It’s the first visible win that makes the next ten feel inevitable.
From there we would build Q2 (Competitor monitoring) in weeks 2–3 and Q3 (Proposal drafts) over weeks 3–5. By the end of the first month, roughly 138 of the 340 wasted hours would be gone.
Book the free audit. 48 hours after our call, a document like this — your workflows, your costs, your numbers — lands in your inbox.
Book Your Free Audit