Industries / Professional services
For professional services firms
Keep the expertise human. Systemise everything around it.
Your firm sells judgment. Around that judgment sits a week of repeated information work: research, intake, briefs, searching for the last version, reports, onboarding, follow-ups. Dekundy Group builds AI-assisted research, intake, knowledge, reporting and client workflows around the work your team repeats, and leaves the professional decisions with the professionals.
Not sure which of these applies to you? The CA$49 audit works that out first, and says so plainly when the answer is that nothing needs building. See the system
- Consultancies
- Agencies
- Accounting practices
- Legal practices
- Recruiting firms
- Business-services firms
- Client intake
- Research agent
- Knowledge assistant
- Proposal preparation
- Recurring reporting
- Document intake
- Project command center
Recognisable problems
Where the week actually goes.
Research starts from zero every time
A new client, a new matter, a new pitch. Someone opens the same tabs, copies the same facts and writes notes nobody will find next month.
Onboarding is reinvented per client
The engagement letter, the questions, the folder structure and the kickoff tasks depend on who set the client up and how busy they were.
The answer exists, somewhere
The policy, the precedent, the template, the rate. It is in a document, an email or a partner's head, and the question gets asked again.
Reports are rebuilt by hand
Monthly client updates, status reports and internal reviews are assembled from timesheets and spreadsheets on the last day of the month.
Featured system
Client intake and onboarding system
A structured request captures what the client can tell you. The agent classifies the service, writes the internal summary and checks for missing information. Automation creates the client record, the project, the task template and the owner. A welcome message with any outstanding questions is drafted for approval, and the onboarding board shows who is blocked and who is ready.
Intake form
Submitted- Company
- Bramble & Co. Accounting
- Service
- Bookkeeping, monthly
- Entities
- 2
- Start
- 1 October
AI classified as: Standard onboarding, tier 2
Onboarding steps
Automation- CRM account created
- Engagement letter generated
- Welcome email sent from template
- Tasks assigned to onboarding lead
- Fee schedule confirmationReview
- Kick-off call scheduling
Components
- Intake form
- AI classification and summary
- Missing-information check
- Client and project records
- Task templates
- Owner assignment
- Welcome message with approval
- Onboarding board
Input
01Client submits information
A structured form captures the essentials once.
AI agent
02AI structures the request
Service type, summary and priority in the firm's terms.
Business logic
03Missing-information check
Required items per service type; gaps become questions.
AI agent
04Internal summary
What the team needs to know before the first call.
Automation
05Owner and project
Records created, owner assigned by service line and load.
Automation
06Tasks
The onboarding template for that service, generated.
Human approval
07Approval
The owner reviews the summary, tasks and message.
Output
08Client communication
Welcome, next steps and the questions, sent once approved.
- Input
- AI agent
- Business logic
- Automation
- Human approval
- Output
Second system
Research and knowledge system
Approved sources are monitored, findings are structured and stored per company or matter, and briefings are drafted with every claim linked to its source. Source selection, permissions and the final judgment stay with your people.
Weekly intelligence briefing
Mon 07:00What changed
- Two competitors changed pricing on entry plans. 3 sources
- New tender published in the target region, closes in 18 days. 1 source
- Prospect account announced a new operations lead. 2 sources
Sources
Agent
Running- Compared against last week
- Drafting summary
Components
- Source registry
- Research agent
- Structured findings
- Human review
- Briefings
- Internal search
Input
01Sources
Only sources your firm approves, with permissions respected.
AI agent
02Agent
Extracts developments and compares them with what is known.
Business logic
03Structured findings
Filed against the client, matter or account.
Human approval
04Human review
A professional reviews before anything is circulated.
Output
05Briefing
Decision-ready, with sources attached.
- Input
- AI agent
- Business logic
- Automation
- Human approval
- Output
What can be built
Agents, automation, dashboards and software for professional services.
AI agents
- Research agent
- Internal knowledge assistant
- Meeting preparation assistant
- Document intake and classification
Automation
- Client intake and onboarding
- Proposal preparation workflow
- Follow-up automation
- Recurring client reporting
Dashboards
- Project command center
- Client status board
- Capacity and deadline view
Software
- Client portal
- Matter or engagement tracker
- Internal request queue
The shape of the change
Typical manual workflow, and an example system architecture.
Illustrative. The left column is how this work commonly runs; the right is one way a system can run it. Neither describes a specific customer, and no improvement is claimed.
Typical manual workflow
- 01
New client details collected
Depends on: An email thread and a partner's notes
- 02
Engagement set up
Depends on: Whoever is free copying last time's folder
- 03
Background research
Depends on: Open tabs and a Word document
- 04
Questions to the client
Depends on: Remembering what was not asked
- 05
Status reporting
Depends on: Month-end assembly from timesheets
- 06
Finding the precedent
Depends on: Asking the person who wrote it
Example system architecture
- 01
Client information captured once through a structured intake
System - 02
Request classified, summarised and checked for gaps
AI - 03
Records, project, tasks and owner created from the template
Automation - 04
Research gathered from approved sources and structured per matter
AI - 05
Professional reviews the summary, the findings and every client message
Person - 06
Reports and briefings assembled on schedule from live records
Automation - 07
Approved knowledge searchable with sources, in one place
System
From published research
The problem is measured.
Published research on information findability and on AI assistance in one customer-support setting. Professional firms were not studied specifically, and these are not Dekundy Group results.
47%
A 2023 Gartner survey of 4,861 employees found 47% struggle to find the information or data they need to do their jobs, while the average desk worker now juggles 11 applications.
Source: Gartner, Digital Worker Survey press release, 10 May 2023 (2023)
+15%
A Stanford and MIT field study of more than 5,000 customer-support agents, published in the Quarterly Journal of Economics, found an AI assistant lifted issues resolved per hour by about 15% on average, with the biggest gains for less-experienced staff.
Source: Brynjolfsson, Li and Raymond (Stanford and MIT), Generative AI at Work, Quarterly Journal of Economics (2025)
Who does what
What the AI handles, what automation handles, what stays with your people.
The AI agent
What the AI actually does
- Classifies and summarises each new client request
- Phrases the missing-information questions
- Gathers and structures research from approved sources
- Answers internal questions with the source attached
- Drafts recurring reports and briefings for review
Automation
What gets automated
- Creates client, engagement and project records
- Generates onboarding tasks from the service template
- Monitors sources on a schedule
- Assembles reports from live records
- Tracks blockers and readiness on one board
Human control
What a person controls
- Owns every professional judgment and every client relationship
- Chooses which sources are approved and who may see what
- Reviews briefings, summaries and reports before they circulate
- Decides scope, fees and anything contractual
Example engagement
Client intake and knowledge system
A realistic scope for this kind of business. Not a fixed product; the exact scope and price are agreed in writing after discovery.
Could include
- Structured intake form
- AI classification and summary
- Missing-information checks
- Client and project records
- Task templates and owner assignment
- Approved-source research agent
- Internal knowledge search
- Recurring reporting
What moves the price
- Number of service types and templates
- Document volume and permissions
- Practice management or CRM tools in use
- Whether a client portal is needed
Pricing
Builds start around CA$2,500
Indicative build budgets CA$3,500 to CA$8,000 · Larger multi-system builds CA$8,000+
Multi-step implementations are quoted according to integrations, channels, data, custom interface, number of workflows and permissions. Not sure where to begin? The CA$49 AI & Efficiency Audit, or a Workflow & System Design from CA$750, comes first.
Adam Dekundy
Founder, Dekundy Group
Built directly with the person who designs it
You are not passed from a salesperson to a delivery team. The work starts by understanding which work your team repeats, defining what better means for the professionals doing it, and building the system around it.
The Build Guarantee
We agree on what the system needs to accomplish before the build begins. If it does not perform the agreed function, Dekundy Group keeps working on the implementation at no additional implementation fee until it does.
Describe one recurring process that consumes more professional time than it should.