You will struggle to quantitatively form the Grignard reagent that you have proposed. The ester functionality will not be stable in refluxing ether. Decomposition is the most likely result. Grignards bearing electronegative atoms are fickle and notoriously difficult to form.

Although it depends on the exact nature of your R groups, I would expect the hypothetical ring opening to occur as depicted, favouring the transdiaxial product. The preference is dictated primarily by the ground state orientation of your R groups. Electronegative R groups may introduce additional effects.

I would try using benzyloxymethyl chloride, or p-methoxybenzyloxymethyl chloride under Barbier type conditions using excess magnesium, zinc, or indium, in the presence of your epoxide, in refluxing THF. Preactivating the metal with 1,2-dibromoethane is recommended.

Unfortunately the benzoic methylene is in the final target. Do you know of a way to oxidize the benzylic position of a BnO to BzO? 

Otherwise I will go with a Barbier/Reformatsky approach as you suggested with zinc metal.

I can't think of any synthetically useful ways of oxidising BnO to BzO, they are all horrible and wouldn't be selective enough to from any useful product as the intermediate tends to hydrolyse and you have another aromatic group that might interfere.

You can give it a try with the benzoyloxymethyl chloride and it may work with Zinc.

The presence of the cyclic acetal complicates the furst plattner rule as the 6-5 bicyclic structure departs somewhat from the chair like transition state. Addition will become difficult as the nucleophile must approach from the concave face of the molecule which is also blocked by your aryl group.

What 1H NMR coupling constant do you observe for the two ring protons vicinal to the amine and aryl group? If it is low (10) then you should be alright.

Regarding the benzyl to benzoyl conversion, I would try radical transformation using Suarez conditions, that is phenyliodonium diacetate/iodine (optional base, such as K2CO3) under visible light irradiation. The initialy formed O-centered radical from the OH group may abstract hydrogen from benzylic position resulting in diodo derivative which will be hydrolyzed during work-up to benzoyl group. The chances for very clean reaction are marginal, but seems to be worth trying from scientific point of view.

I agree with Harris overall. I may suggest also using Samarium II iodide in Barbier type reaction. 

Thank you all for your suggestions, the Samarium Barbier actually looks to be the most promising according to literature reports.

Here are some relevant papers if anyone is interested.

Tetrahedron Letters,Vo1.25,No.30,pp 3225-3226,1984.

J . Am. Chem. Soc. 1987, 109, 4424-4426.

Chem. Soc. Rev., 2013, 42, 9155-9183.

Samarium seems like a good option to make your desired nucleophile however the epoxide might not be electrophilic enough and SmI2 is a powerful reductant. It may be worth adding NiI2 to aid in transmetallation in situ, or potentially a lewis acid to activate the epoxide.

Personally, I would be worried because of the carbenoid character of organometallic reagents of the type ROCH2M. Perhaps I would consider the possibility of running a Reformatsky reaction using PhC(O)CH2X followed by a Baeyer-Villiger oxidation. However, I am not sure whether the latter would proceed with the wanted regioselectivity...

I have successfully worked with the lithiated benzyl ether.  (W.C. Still, JACS 100:1481 (1978).  If you can use that, or the p-methoxy analogue, you'll probably be in business.

If you can make PhCO2CH2-SnBu3, it might be possible at -78ºC (or -100ºC) to transmetalate with BuLi, but the reagent would not be stable for long even at those temperatures.  I'd also give t-BuLi a shot, in the presence of the epoxide. The cuprates are worth trying in both cases.

The issue, with any of the proposed organometallic reagents, is competition between the two electrophiles, benzoate and epoxide. 

Thank you all for your quick responses, 

Eric, I'll consider the NiI2, and/or Lewis acids.

Yvan, I like the idea of BV oxidation, the cloroacetophenone would be more predictable like a Reformatsky nucleophile.

James, unfortunately the 4-MeO-version is not so much an option based on the target molecule. The possibility of the BzOCH2Cl attacking itself seems to be the main crux of the issue, but from what I've read, organosamarium reagents tend to not and do not undergo syn elimination. However your advice is greatly taken, as it is in line with Eric's suggestion of an accompanying Lewis acid or transmetallation agent.

Thanks again.

I would rather look at  PivOCH2ZnI instead; with chloride you will not get any insertion of Zn metal. Paul Knochel in 1990-2000 did a lot with this reagent. Epoxide ring-opening might work in the presence of Cu(I)+BF3-Et2O or smth alike. You may need screen Cu(I) reagents to find the most suitable for this reaction.Possibly, BzOCH2ZnI can be done and used as well. As for BzOCH2MgCl, I have serious doubts it is possible to prepare and use it with any reasonable yields.

The Baeyer Villiger is a nice idea but it would give the opposite regioselectivity resulting in the phenol ester.

If you are making the iodide Finkelstein chemistry works well, but use it fresh, handle it quickly and avoid light.

Lithium halogen exchange of BzOCH2I or BzOCH2Br with 2 e.q. of t-BuLi at -78 or -100 °C in the presence of CuI may also be worth looking into.