Questions before the first agent build.

Alderson builds AI agents for experiential teams. We clean the project data behind one repeated job, build the agent, train the first users, and measure whether the work gets easier.

Experiential agencies, event teams, production teams, field marketing groups, and operators with repeated deadline work across pitches, recaps, estimates, production planning, field reporting, or campaign intelligence.

One repeated job gives the work a real owner, a clear data set, a baseline, and a result the team can judge. Broad AI programs usually fail because the data is messy and no one knows what output should be trusted.

Common first agents include pitch proof, recap assembly, estimate precedent, production planning, field reporting, sensory research, and campaign intelligence.

It depends on the job. Typical material includes final decks, recaps, budgets, scopes, bids, actuals, run sheets, vendor notes, approvals, photos, event results, field notes, surveys, and client feedback.

Do not send passwords, API keys, access tokens, banking details, raw HR, payroll, or medical data, regulated personal information, or restricted client material through public email or website channels. Sensitive access should be scoped and handled through an approved process.

We work with the owner to separate finals from drafts, identify authoritative folders and systems, remove stale or risky material, document source-of-truth rules, and create sample scenarios for human review.

No. Alderson agents are built to support human teams. Outputs still need review before client-facing, financial, legal, HR, procurement, safety, or other high-impact use.

We pick one baseline before the build. That might be manual search time, drafting time, rework, turnaround time, approval friction, or adoption. The goal is to measure whether the job became easier, not to promise a generic savings number.

The timeline depends on workflow scope, source readiness, access, stakeholder availability, and procurement or legal review. A diagnostic or pre-audit can be shorter; a first workflow build needs enough time for source prep, workflow design, build, QA, training, and measurement.

Not by default. We prefer approved exports, scoped folders, approved samples, or named least-privilege accounts. Broad production access, admin access, APIs, or service accounts require explicit scope and security review.

Usually the first pass works around the tools the team already uses: Drive, SharePoint, Slack, Teams, Sheets, Excel, Asana, Monday, Cvent, Splash, HubSpot, asset libraries, and similar systems. Exact integrations depend on access, APIs, security requirements, and the selected workflow.

Alderson may use AI-enabled tools to support analysis, drafting, coding, data cleanup, retrieval, evaluation, and agent implementation. Client restrictions on AI tooling should be documented before project data is shared. Final commitments belong in the SOW or DPA.

The intended operating position is that raw client confidential material should not be reused to train generalized Alderson models or submitted to consumer or public AI services where inputs may be used to train models unless the client approves in writing. Final language should be set in the governing agreement.

Ownership and usage rights should be defined in the MSA or SOW. The default commercial intent is to make sure the client can use the paid-for deliverables for the scoped workflow while Alderson preserves non-client-specific methods, templates, and know-how. Counsel should review final language.

There are three paths: maintain the current agent, stabilize data quality and adoption, or expand to the next job with a clear return case. Renewal should be based on evidence from the first build, not a generic roadmap.

Alderson is not a generic prompt-training workshop. We start with the work: who owns it, where the data lives, what output is needed, what can be trusted, what must be reviewed, and how the result will be measured.

Send one repeated deadline job to hello@alderson.ai. Keep the first note simple: the job, team roles involved, current tools, where the data lives, and what would make the work easier. Do not include secrets or restricted client material in the first message.