Part 1 The problemWhy teams need this
01 · THE PROBLEM
AI that writes without sources fills the gaps with confident nonsense, and that's what gets a brand embarrassed. Doing the research as its own step means every draft is built on material that actually exists.
Part 2 How it worksWhat it does, step by step
02 · WHAT IT DOES
It picks up a new brief, locks it so nothing else touches it, and searches the web for credible sources on the topic. The research is attached to the post for the writer. If the search fails, the post is flagged instead of moving on half-empty.
03 · HOW IT RUNS
Step by step, as built.
Find work
Fetches one post waiting for research. No work, and the run ends quietly.
Lock it
Marks the record as in progress before doing anything slow.
Search
Exa returns sources for the topic, which are assembled into a research brief.
Save or fail loudly
Research is written back to the record; an Exa failure routes to an error status.
04 · TOOLS AND APPS
Built around the systems already in the process.
05 · WHEN SOMETHING BREAKS
Failure is designed in.
- 1node has its own error branch, so a failure is routed and handled, not just logged.
- 1decision point (IF or Switch) checks the data before it moves on.
- ✓Anything unhandled triggers our central error workflow, so a crash gets reported instead of failing silently.
Standard on every build
- Schema validation before downstream writes
- Retry and error routes for external API failures
- Duplicate-safe processing and idempotent updates
- Human approval where the action carries business risk
- Execution logging for support and audit review
Part 3 The impactWhat it's worth, and how we'd build yours
06 · PROJECTED IMPACT
What it should change in the business.
Projections for a typical deployment. The calculation below shows the math, and you can put in your own numbers.
07 · ROI CALCULATION
How it pays back in your business.
The starting numbers are a hypothetical deployment sized to the projections above. Change any of them to your own volumes and costs, and the math updates underneath.
- Full-time equivalent freed
- 0.1 people
- Gross value
- $6,528
- Running cost
- −$180
- Return per $1 of running cost
- $36.3
The math: 8 posts researched × 120 min × 12 months × 85% ÷ 60 = 163 hours a year × $40/hour = $6,528. Net value = gross value − $180 running cost a year.
08 · HOW WE'D BUILD YOURS
How we'd build yours.
- Discover: map the current process, systems, volumes, owners and exceptions.
- Design: define the canonical data model, approvals, retries and system boundaries.
- Build: implement credentials, nodes, validation and observable error routes.
- Prove: run controlled data through success, duplicate and failure scenarios.
- Operate: publish runbooks, ownership and measurable service levels.
