Choosing an AI provider
Should our ERP partner do our AI quoting and order entry, or should we use an AI consultancy?
Last updated
Use your ERP partner when the work stays inside the ERP: configuration, upgrades and turning on the ERP's own AI features. Use an AI consultancy when quoting or order entry spans email, documents, field service and the ERP, and needs a measured baseline and accuracy proven in shadow mode. Most organisations use both, with the ERP partner owning the ERP and the consultancy building the AI layer.
Who this is for
This page is for application and ERP managers, IT leads and operations managers at mid-sized to enterprise organisations, often multi-site with no in-house AI team, running Pronto Xi, Business Central, Dynamics 365, NetSuite, SAP or SAP Business One, where emailed RFQs, PDF orders or field service jobs are retyped into the ERP.
Solve8 is an AI consultancy and delivery firm working beside ERP partners in Sydney, Brisbane, Melbourne, Perth and across Australia and New Zealand, in fixed-fee phases: discover one process with a measured baseline, prove it in shadow mode, build it on your infrastructure in Australia, then run it or hand over the code.
Who fits which job
| Work | ERP partner | AI consultancy | Internal AI hire |
|---|---|---|---|
| Turning on AI features your ERP vendor ships | Best fit | Can advise on fit | Rarely needed |
| ERP configuration, upgrades and custom fields | Best fit | Not their role | Not their role |
| Quotes or orders built from email, PDFs and field service data | Possible if the partner has AI staff | Best fit | Possible with senior experience |
| Custom approval rules, price checks and exception queues | Inside the ERP, yes | Across systems, yes | Possible |
| Measuring a baseline and proving accuracy in shadow mode | Not usually offered | Core of the work | Possible once the person is in place |
| Many AI use cases over several years | Partly | Builds the first ones | Best fit long term |
When your ERP partner is enough
The feature already exists in the ERP
If your ERP vendor ships an AI feature that fits, such as Business Central's Sales Order Agent, your ERP partner is the natural choice.
The work stays inside one system
Mappings, workflows and reports inside the ERP are partner work.
Your data is clean
Native agents depend on good item, customer and contact data. Where that is in order, configuring them is often quicker than a build, unless an ERP upgrade is about to change things.
When an AI consultancy fits
The process spans several systems
Quoting that pulls defect records from simPRO, AroFlo or Uptick, pricing from the ERP and requests from email crosses boundaries no ERP partner owns.
Steps are document heavy
RFQs, drawings and PDF orders in many layouts need extraction, matching and checks before anything reaches the ERP.
Your rules are your own
Price tolerances, margin floors and branch credit rules usually need logic outside the native feature.
You need proof before you commit
Measure today's turnaround and error rate, then run the AI in shadow mode on live work until it meets your accuracy bar. Solve8's defect quoting AI for a national fire protection provider, built in the client's own cloud in Sydney, is entering shadow mode this way.
When to hire an AI engineer instead
Hire after the first use case
Once AI is a standing program with several use cases, hire. A common route is a consultancy delivering the first process with full code and documentation, then a hire to run and extend it.
How the ERP partner and an AI consultancy work together
Agree who owns what
The partner owns the ERP and upgrades. The consultancy owns the AI layer and review screen.
Agree the write-back
Which records the AI creates, through which API and integration user.
Build and shadow
The consultancy builds against the partner's test company, then the AI drafts on live work without posting, measured field by field.
Go live with approvals
A person approves each record before it posts.
Plan for upgrades
Retest the integration before each ERP upgrade.
Ten questions to ask any provider
Which ERP write-backs have you built?
Name the ERP, version and record type, and show one working.
Where does our data sit?
Where emails, documents, model calls, backups and support access sit, including offshore.
Is our data used to train any model?
In writing, for the provider and every service it uses.
How is a record approved before it posts?
Who approves, on what screen, and what is logged.
How will you measure accuracy?
On our documents, against a baseline, with a bar we set before go-live.
What happens when the AI is unsure or wrong?
Which checks hold a record back and how it is paused.
Which of our systems will you connect?
Every system and who owns each connection.
Who supports it after go-live?
Who monitors accuracy and retests after ERP upgrades.
Can we keep the code?
Code, prompts, configuration and documentation, in our environment with our admin access.
What do we keep if we stop early?
What we hold after phase one that someone else could pick up.
Australian specifics
The Australian Government's Questions to ask AI suppliers guide on ai.gov.au covers risk, transparency, accountability, data, testing, human oversight and fairness, and says what to capture in writing, including where data is stored or processed and whether it will train AI models. Send it to every shortlisted provider, ERP partner included. Quotes and orders carry customer contact details, so sending them offshore can engage APP 8 of the Privacy Act, and the Privacy Act 2020 applies in New Zealand.
Who should own it
ERP or application manager
Owns the partner relationship and the integration user.
Operations or sales manager
Owns the process, exceptions and the accuracy bar.
IT lead
Owns hosting, security and data location.
Related work
AI quoting for fire protectionDefect records categorised and quote lines suggested, built in the client's cloud in Sydney.
Purchase orders into the ERPHow AI reads PDF orders, checks them and writes them back.
AI agentsAgents that work across email, documents and business systems with approvals.
AI business case toolkitBaseline and business case templates to take to your executive.
Sources
Australian Government, ai.gov.au, Questions to ask AI suppliers: https://www.ai.gov.au/sites/default/files/2026-07/Questions-to-ask-AI-suppliers.pdf
Microsoft Learn, Sales Order Agent overview for Business Central: https://learn.microsoft.com/en-au/Dynamics365/business-central/sales-order-agent
Questions
Can our Pronto, Business Central or NetSuite partner build AI quoting themselves?
Some ERP partners can, especially where the ERP vendor ships an AI feature that fits, such as the Business Central Sales Order Agent. Where quoting draws on email, PDFs and a field service platform as well as the ERP, ask the partner which AI write-backs they have delivered and how they measured accuracy before go-live.
Should a 600-person company with no AI team hire an AI engineer or use a consultancy?
A 600-person company with no AI team usually gets its first use case live sooner through a consultancy, with code and documentation handed over. Hiring makes sense once AI is a standing program with several use cases, and the new hire then starts with a working system to run and extend.
Will our ERP partner support an AI layer built by another firm?
Most ERP partners will support the ERP while another firm supports the AI layer, provided the boundary is agreed early. Settle which records the AI writes, which API and integration user it uses, and who retests after ERP upgrades. Put those points in writing with both parties before the build starts.
Three AI vendors pitched us the same thing. How do we tell them apart?
Ask each vendor to show a working write-back into your ERP type and to explain how they would measure accuracy on your own documents before go-live. Send all three the same written questions, including the ai.gov.au supplier questions on data location and training, and compare the answers line by line.
How do we keep AI quoting and order data in Australia?
Ask every provider where emails, documents, model calls, logs and backups are stored and processed, and get the answer in writing. The simplest way to keep data onshore is to run the AI layer in your own cloud environment in an Australian region, with model services chosen for Australian data processing.
Can we move support of an AI build to our ERP partner or an internal team later?
Yes, if the contract gives you the code, prompts, configuration and documentation, and the system runs in your own environment with your admin access. Ask for a handover session and a written runbook covering monitoring, exception handling and retesting after ERP upgrades, so another team can take over support.
Related
- We get quote requests by email. Can AI match the SKUs, apply customer pricing and create the quote in our ERP?How distributors use AI to read emailed RFQs, match SKUs, apply customer pricing and margin rules, check stock and draft quotes in Business Central.
- AI agents for Australian organisations, built inside your systemsAn AI agent is software that reads incoming work, decides within your rules and acts across your systems. A person approves the result.
- System integrationConnections between the systems you run.
- AI business case toolkitA one-page business case, a vendor checklist and the security questions IT will ask, so you can take an AI pilot to your COO, CFO and CIO. Copy it, print it or forward it.
- Customers email us PDF purchase orders and staff retype them into the ERP. Can AI do this reliably?How AI reads emailed PDF purchase orders, checks them against customer, item and price data, routes exceptions to a person, and writes orders to your ERP.
- How to choose an AI consultancy in AustraliaA fair guide to the four types of AI firm, how they compare, twelve questions to ask and the red flags to watch for.
Have this problem across your branches?
A 30-minute review with a senior consultant. You leave with a written one-page summary you can share internally.
Fixed-fee phases, agreed before we start. Stop after any phase.