SmartWork — Working AI workflows inside your business in 4 weeks

Packaged AI software has fixed features and teams end up shaping their work around the tool instead of the tool fitting the work. Agentic workflow tools are different. They read documents, run multi-step tasks, operate software on your computer and are configurable to any workflow your team owns.

We run our own company on Claude — generating proposals, websites, strategy, sales, support, contracts. SmartWork puts the same operating model inside yours.

In four weeks, we provide two workflows automated on your real data, AI champions trained to build with Claude and a playbook to extend the pattern across the firm.

What a SmartWork sprint delivers.

Working agents and workflows configured to your business. Examples below are drawn from live engagements.

you: Generate a research brief on Acme Corp before tomorrow's meeting
Loading public filings, news, LinkedIn data...

EXTRACT
• Revenue, headcount, geographic footprint
• Recent funding rounds and leadership changes
• Top 5 customers and competitors mentioned

CALCULATE
• 3-yr revenue CAGR: 24%
• Engineering hiring: +18% YoY
• Tech stack: 4 AI vendors detected

DRAFT
• 1-page executive summary
• 5-section deep brief in your template
• Talking points for the meeting

OUTPUT
✓ /briefs/acme-corp-2026-05-03.pdf
✓ Source citations on every claim

Turnaround: 4 minutes. Manual baseline: 3-4 hours.

Report Generation

Read the data. Run the numbers. Write the report.

For workflows where someone reads documents or queries databases, pulls structured fields, runs calculations, and writes a polished report. Common examples: account research, supplier intelligence, competitive analysis, weekly business reviews.

  • Uses SmartDoc to extract information documents (e.g spreadsheets)
  • Pulls fields from your databases and spreadsheets
  • Runs mathematical calculations using deterministic code — so the numbers are exact every time
  • Writes output into your existing templates (PDF, Word, Excel)
you: Draft a 4-week pilot proposal for Horizon Logistics
Loading customer profile... industry: shipping

CUSTOMIZE
• Pulled 3 past meeting notes from /Notes/
• Adapted scope from past logistics customers
• Inserted shipping case studies (3)
• Priced at standard pilot tier

DRAFT
✓ Cover letter, scope, timeline
✓ Pricing table with split payment terms
✓ Team bios and case study appendix
✓ All sections in OphieAI house style

OUTPUT
✓ /proposals/horizon-logistics-2026-05-03.pdf
✓ Mirror deck for the meeting (.pptx)

Turnaround: 6 minutes. Manual baseline: half a day.

Proposal Generation

Tailored proposals drafted in minutes, not days.

For sales and BD teams who today spend half a day per proposal — pulling boilerplate, customizing scope, pricing the work, and assembling the deck. The agent reads the customer's context, adapts each section, and outputs a branded document ready for review.

  • Reads the customer profile, past meetings, and prior proposals
  • Uses SmartDoc to read customer-supplied RFPs and briefs to pull out scope and requirements
  • Adapts scope, pricing, and case studies to the customer's industry
  • Outputs a branded PDF and mirror deck in your house style
you: Triage my inbox
Scanning 127 emails from the last 48 hours...

PRIORITY — ACTION REQUIRED (3)
• Apex Construction — Contract amendment due Friday
• Horizon Logistics — Pricing response requested
• Internal — Board deck review by Thursday

CUSTOMERS (5)
• Apex — contract amendment attached
• Horizon — Q2 expansion pricing
• Meridian — Tuesday confirmation
• +2 more

FOLLOW-UPS PENDING (4)
⚠ Legal review at Apex (5 days unanswered)
⚠ Payment terms at Horizon (5 days)

Filtered: 112 newsletters, system notifications, social

Email Intelligence

Your inbox triaged. Action items surfaced.

Reads your inbox, classifies senders, surfaces what actually needs you, and tracks pending follow-ups. The morning starts with a 30-second view of priority items, with newsletters, system notifications, and automated noise already filtered out.

  • Categorizes by sender type: Customer, Prospect, Partner, Internal
  • Surfaces action-required emails with deadlines
  • Tracks pending follow-ups and response wait times
you: Show me how the agent prices a custom proposal
Loading workflow rules...

═══ /skills/proposal-pricing/RULES.md ═══

# Custom Proposal Pricing
1. Match scope to standard tiers first.
2. Custom scope: cost bottom-up, then mark up
1.4x for fixed-fee delivery.
3. Always offer 50/50 split on engagements
above $50K (signature + delivery).
4. Government customers: discount 12%, model
60-day payment terms.
5. If the deal affects a reference account,
escalate to a partner before sending.

Decision logic that lived in Alex's head for
three years. Now every champion applies it
the same way.

Codified Knowledge

The rules in your senior people's heads, written down and enforced.

"Always offer 50/50 split on engagements above $50K." "Discount 12% for government customers, never below cost." Tribal decision logic gets captured as structured rules that the agent enforces and new hires can read on day one.

  • Business rules written as plain text — anyone can read or edit them
  • Your team adds new rules whenever a new situation comes up
  • Stored in your own files, with a full history of every change
  • The knowledge stays even when senior people leave
you: Show me my SmartWork setup
Loading Claude environment...

═══ YOUR REPOSITORY ═══

📄 CLAUDE.md — working conventions, voice, rules

📁 skills/
├── account-research/
├── proposal-generator/
├── email-intelligence/
└── pricing-rules/

📁 knowledge-base/
├── customers/
├── templates/
└── case-studies/

📁 connectors/
├── sharepoint
├── outlook
└── smartdoc

All files on your machine, owned by you.

Claude Environment

Your files, your skills, your knowledge base.

A Claude Code or Claude Cowork environment set up around your templates, your work types, and the way your team operates. Your skills, your business rules, your knowledge base — all on your machine, owned by you.

  • A central instructions file that tells Claude how your team works
  • A library of skills, each one built for a workflow you chose
  • Direct connections to SharePoint, Outlook, and your other systems
  • SmartDoc plugged in for reading PDFs
champion: Add a new workflow for monthly board reporting
Setting up new workflow...

ADDING TO THE SYSTEM
✓ /skills/board-report/ created
✓ Reused reading prompts from /account-research/
✓ Pulled report template from /knowledge-base/
✓ First test: 4 of 6 sections filled correctly
✓ 2 tricky cases logged for fixing

TIME TAKEN
Adding the next workflow: 2 days
Adding the first one (with us): 4 weeks

Each workflow after the first is faster
than the last — the work you've already
done carries forward.

Trained Champions

AI champions fluent in Claude to drive the change

AI champions from your team pair with us across the four weeks. They co-design the first workflow, run it on live work, and leave with the skill to extend the pattern themselves. Each subsequent workflow is faster than the last.

  • AI champions trained inside Claude on real workflows
  • Step-by-step guides for each workflow — how to run it, what to check, when to ask for help
  • A guide for your team on how to add the next workflow themselves

The engine

Powered by Claude.

SmartWork runs on Anthropic's Claude Code and Cowork — the same engines we use to run our own company.

  • AI that takes action. Claude reads, decides, runs the work, and operates other software in your environment — not just chat.
  • Your business, your context. SmartWork sets up a skill library configured to your customers, projects, templates, and rules — stored on your machine, controlled by you.
  • The most capable AI available today. The same Claude that the world's most demanding engineering teams trust for complex, multi-step work.

The approach

A fixed-scope sprint, not an open-ended project.

Actual AI-powered workflows running in your environment in four weeks

Week 0 Readiness

Identify target workflows and get the inputs ready — past projects, documents, templates, and confirmed champions all in place before Week 1 starts.

Week 1 Configure & build

Set up Claude environment, connections, and the first version of chosen workflows.

Week 2 Live client work

Your champions run the workflows on live client projects. We work alongside them, fixing the tricky cases that only show up on real work.

Week 3 Handover & next steps

Debrief on what worked, what didn't, and a plan for scaling the approach across the rest of the firm.

The deliverables

What you walk away with.

  1. 1

    Selected workflows running on AI.

    A working automation against your real data that delivers real manhour savings.

  2. 2

    A team of AI builders.

    The people who run the workflow learn to build on it themselves inside Claude. They leave the sprint with the skill to ship the next workflow without us.

  3. 3

    The rules in your senior people's heads, written down.

    Decision logic that previously lived in someone's head gets captured as written rules the agent follows and new hires can read on day one.

  4. 4

    A playbook to extend the pattern.

    The setup, skills, and rules from the sprint become the template for the next workflow. Each subsequent sprint is faster than the last — the work you have already done carries forward.

Bring a workflow that is taking more of your team's time than it should. We will run a SmartWork sprint on it and leave behind something they can extend.

Book a sprint