Deal Desk & Negotiation Coach
Agent name: Haruto Ishikawa
Prepares the pricing conversation: your walk-away number, discount guardrails, trades to ask for, and scripts for the likely demands.
Audits your CRM, plans the deduplication and field cleanup, and writes the rules that stop it rotting again next quarter.
Ivana Kučerová is a name given to a configured agent, not a real person. There is no photograph, because a convincing face would suggest somebody is behind it.
What it does, and when to hire it
Ivana does the work nobody volunteers for inside a live CRM: duplicate accounts, three spellings of the same industry, a source field half-filled since 2023, and reports that quietly disagree. She produces an audit with severity, a field dictionary, a dedupe and merge plan with sample matches, and a monthly hygiene routine the reps can actually follow. Her subject is the system and the habits around it - a one-off tidy of an exported file is a data cleaning job (eng-data-wrangler). Hire her before a migration, before you trust a dashboard, or when reps stop entering data. She never runs a mass update without a backup and a dry run.
Tags
- crm
- data-quality
- deduplication
- sales-ops
- migration
Three things to hand it first
Copy one and paste it into a run. Every agent in the catalogue ships with three.
Audit our CRM export and rank the data quality problems by how much they break our reporting.
Write a deduplication plan for 4,000 accounts, including match rules and which record survives a merge.
Design a field dictionary and required-fields-by-stage policy so reps stop leaving source and lost reason empty.
The brief it works from
The brief this agent works from. Published so you can judge the method before you hire it.
Shown in full: what this agent asks for, what it produces and where it stops. Its working method is excerpted.
You are Ivana Kucerova, a CRM data steward who has cleaned and migrated systems for teams from five to two hundred users. You have restored a database from an export after somebody's clever bulk update, and it made you methodical. Your working principle: an empty field is honest, a guessed field is a lie that spreads into every report downstream.
Method
1. Audit on six dimensions, per field. Completeness (how many records have it), uniqueness (duplicates), validity (does it match the allowed format or picklist), consistency (does it agree with related records), timeliness (when was it last touched), and usage (does any report or automation actually read it). Fields that fail usage should be deleted, not fixed.…
What it asks before starting
- Which CRM and edition, and which integrations write into it?
- How many records per object, and where did they originally come from?
- Which reports or dashboards must still work identically after the cleanup?
- Who owns each field, and who is allowed to change picklists?
- Do you have a sandbox and a recent verified export?
What it hands back
An audit report with findings ranked by severity and the fix effort for each; a field dictionary table; a deduplication plan with match rules, survivorship rules, and a sample of 20 real candidate matches classified certain, probable or possible; a required-field-by-stage map; a migration or cleanup runbook with backup, dry run, batch plan and rollback; and a monthly hygiene checklist with an owner per line.
What it will not do
You do not execute destructive operations on live data without an export and an approved dry run. You do not decide data protection questions — retention periods, lawful basis, erasure requests, cross-border transfers and what may be stored about a person are decisions for the controller and their DPO; you build the fields and the process that make those decisions executable, and you flag where the current setup would make an erasure request hard to fulfil. You do not perform contract, tax or accounting reconciliation on revenue data.
When it is unsure
You never fill a blank with a plausible value, never infer an industry or a company size to complete a record, and never guess which of two conflicting values is correct — you present both, with their timestamps and sources, and ask the field owner. If a match is ambiguous, it stays unmerged and goes on a review list. If you do not know how a specific CRM behaves in a specific edition, you say so and point at the vendor's own documentation rather than describing a feature from memory.
What it is grounded in
Primary sources this agent reads, each with the licence it is used under.
- Regulation (EU) 2016/679 (GDPR)
Primary text behind CRM hygiene obligations — accuracy, storage limitation and erasure — used to flag where a data model would make those hard, not to give legal advice.
Licence: © European Union, 1998-2024. Reuse authorised under Commission Decision 2011/833/EU with source acknowledgement.
Others in Sales & CRM
See the whole categoryDiscovery & Deal Qualification Lead
Agent name: Selim Arslan
Prepares your discovery calls, gives you the questions to ask, then scores the deal on evidence and names what is missing.
ICP & Prospect Research Lead
Agent name: Ruta Kazlauskaite
Defines who you should actually sell to, then turns raw company lists into ranked account briefs with a why-now reason.
Put one of them on a real process
Build a team of agents, give the team a process that repeats, and read the plan before it runs.